TECH & SPACE
🇬🇧
Povratak
🇬🇧 EN
Umjetna inteligencijadb#2549

Claude Code sada preuzima vaš kompjuter: koliko je to opasno?

(1d ago)
San Francisco, United States
arstechnica.com
Claude Code sada preuzima vaš kompjuter: koliko je to opasno?

Claude Code sada preuzima vaš kompjuter: koliko je to opasno?📷 © Tech&Space

  • ★Claude Pro i Max dobivaju pristup radnoj površini
  • ★Anthropic upozorava na 'nije apsolutno' zaštitne mjere
  • ★Tehničke i sigurnosne nedoumice ostaju nerazjašnjene

Anthropic je upravo stavio svoje AI agente na radnu površinu korisnika — doslovno. Claude Code, dio Claude 3.0 Sonnet paketa za Claude Pro i Max pretplatnike na MacOS-u, sada može widowati po ekranu, klikati po ikonicama i izvršavati zadatke kao da sjedi za stolom umjesto vas.

Funkcija je označena kao 'research preview', što u prijevodu znači: 'može raditi, ali ništa ne garantiramo'. I dok tehnički demo impresionira — agent koji samostalno pretražuje direktorije, pokreće skripte ili čak instalira softver — sigurnosna upozorenja Anthropica zvuče sve manje kao oprez, a više kao priznanje. 'Safeguards aren’t absolute', stoji u službenoj dokumentaciji, što je blagi način reći: 'ako nešto pođe po zlu, neće biti naš problem'.

Ono što je zaista novo nije samo mogućnost automatizacije, već dubina integracije. Dok GitHub Copilot ostaje unutar uređivača koda, a AutoGen u chat sandboxu, Claude Code prelazi granicu između alata i agenta.

Što se događa kad AI agentima damo ključeve od naših računala — a oni još uvijek griješe

Što se događa kad AI agentima damo ključeve od naših računala — a oni još uvijek griješe📷 © Tech&Space

Što se događa kad AI agentima damo ključeve od naših računala — a oni još uvijek griješe

To nije više pomoćnik — to je korisnik s pristupom vašem računalu, i to s istim ograničenjima kao i vi: može griješiti, može ne razumjeti kontekst, može se zaglaviti u petlji. Problem je što nismo dobili odgovore na ključna pitanja.

Koliko je ovaj agent izoliran? Tko odobrava svaki korak?

Što se događa ako agent odluči 'optimizirati' zadatak na način koji korisnik nije predvidio — recimo, obrisati direktorij umjesto da ga organizira? Anthropic samo kaže da će 'sometimes take much longer and be more error-prone', što zvuči kao tehnički način reći: 'ne računajte na to'.

U praksi, Claude Code trenutačno najviše podsjeća na beta verziju digitalnog asistenta: koristan za demonstracije, ali nepouzdan za stvarni rad. To ne znači da je koncept loš — slični alati već postoje u Enterprise okruženjima, od Microsoftovih Power Automate do Googleovih AI agenata.

U svakom slučaju, Claude Code predstavlja važan korak u razvoju AI tehnologije i njene integracije u našu svakodnevnu rutinu. Kako će se ova tehnologija dalje razvijati i kakve će prednosti i rizike donijeti, samo će vrijeme pokazati. Jedno je sigurno: budućnost je ovdje, i ona je inteligentna.

ClaudeAI AgentsKeyboard SecurityAI Error Rate

//Comments

Uredi u foto-review →
AIAmazon’s $50B OpenAI bet: Trainium’s real test begins nowSpaceMapping the Local Bubble’s magnetic field reshapes cosmic scienceAIGoogle’s Gemini games flop: AI hype hits gamer realitySpaceStarship’s Tenth Test: The Reusability Threshold CrossedAINvidia’s AI tax: half your salary or half your careerSpaceJWST peels back dust to reveal star birth in W51AITriangle Health’s $4M AI won’t replace your doctor—yetSpaceAI’s Copyright Chaos Threatens Space Exploration DataAIHumble AI is just healthcare’s latest buzzword for ‘don’t trust us yet’SpaceExoplanet spins confirm a planetary mass ruleAIOpenAI’s teen safety tools: open source or open question?GamingCrimson Desert’s AI art fail: a mockup that slipped throughAITinder’s AI gambit: swiping left on endless swipingGamingPearl Abyss hid AI assets in Crimson Desert—now players want answersAINVIDIA’s Alpamayo AI: Self-Driving’s Hardest Problem or Just Another Demo?GamingCapcom Rejects AI AssetsAIWaymo’s police problem exposes AV’s real-world blind spotsRoboticsAtlas Redefines Humanoid DesignAILittlebird’s $11M bet: AI that reads your screen—without the screenshotsRoboticsOne antenna, two worlds: robot sniffs out realityAIUK firms drown in AI hype, emerge with empty spreadsheetsRoboticsDrone swarms take flight—but not off the demo lot yetAIApple’s Gemini Distillation: On-Device AI Without the Cloud HypeTechnologyTaiwan’s chip giants bet on helium and nukes to dodge supply shocksAICapcom’s AI partner talk is just corporate speak for ‘we’ll use it carefully’TechnologySignal’s phishing crisis exposes the limits of encrypted trustAIOpenSeeker’s open gambit: Can 11K data points break AI’s data monopoly?MedicineTelmisartan Boosts Cancer TreatmentAIGimlet Labs Solves AI BottleneckMedicineXaira Unveils X-CellAIHelion Powers OpenAIMedicineAI Fails to Speed Lung Cancer DiagnosisAINVIDIA’s OpenShell: Security for AI Agents or Just Another Hype Shell?AIDRAFT Boosts AI SafetyAIProject Glasswing: AI finds flaws everywhere—except in its own hypeAIPAM: Complex Math for a 10% Performance HitAIOpenAI’s erotic chatbot pause exposes AI’s adult content dilemmaAIAI Ranks Recovery Factors—but Who’s Really Listening?AIDeepMind’s AI safety play: real guardrails or just another demo?AILSD for MLLMs: Reinforcement Learning Cuts the Demo FatAIMicrosoft’s 700B AI bet: Hype or a real retail crystal ball?AIAdobe & NVIDIA’s real-time trick shouldn’t work—but it doesAIEmbeddings hit their limits—and no one’s checking the fine printAIAmazon’s $50B OpenAI bet: Trainium’s real test begins nowSpaceMapping the Local Bubble’s magnetic field reshapes cosmic scienceAIGoogle’s Gemini games flop: AI hype hits gamer realitySpaceStarship’s Tenth Test: The Reusability Threshold CrossedAINvidia’s AI tax: half your salary or half your careerSpaceJWST peels back dust to reveal star birth in W51AITriangle Health’s $4M AI won’t replace your doctor—yetSpaceAI’s Copyright Chaos Threatens Space Exploration DataAIHumble AI is just healthcare’s latest buzzword for ‘don’t trust us yet’SpaceExoplanet spins confirm a planetary mass ruleAIOpenAI’s teen safety tools: open source or open question?GamingCrimson Desert’s AI art fail: a mockup that slipped throughAITinder’s AI gambit: swiping left on endless swipingGamingPearl Abyss hid AI assets in Crimson Desert—now players want answersAINVIDIA’s Alpamayo AI: Self-Driving’s Hardest Problem or Just Another Demo?GamingCapcom Rejects AI AssetsAIWaymo’s police problem exposes AV’s real-world blind spotsRoboticsAtlas Redefines Humanoid DesignAILittlebird’s $11M bet: AI that reads your screen—without the screenshotsRoboticsOne antenna, two worlds: robot sniffs out realityAIUK firms drown in AI hype, emerge with empty spreadsheetsRoboticsDrone swarms take flight—but not off the demo lot yetAIApple’s Gemini Distillation: On-Device AI Without the Cloud HypeTechnologyTaiwan’s chip giants bet on helium and nukes to dodge supply shocksAICapcom’s AI partner talk is just corporate speak for ‘we’ll use it carefully’TechnologySignal’s phishing crisis exposes the limits of encrypted trustAIOpenSeeker’s open gambit: Can 11K data points break AI’s data monopoly?MedicineTelmisartan Boosts Cancer TreatmentAIGimlet Labs Solves AI BottleneckMedicineXaira Unveils X-CellAIHelion Powers OpenAIMedicineAI Fails to Speed Lung Cancer DiagnosisAINVIDIA’s OpenShell: Security for AI Agents or Just Another Hype Shell?AIDRAFT Boosts AI SafetyAIProject Glasswing: AI finds flaws everywhere—except in its own hypeAIPAM: Complex Math for a 10% Performance HitAIOpenAI’s erotic chatbot pause exposes AI’s adult content dilemmaAIAI Ranks Recovery Factors—but Who’s Really Listening?AIDeepMind’s AI safety play: real guardrails or just another demo?AILSD for MLLMs: Reinforcement Learning Cuts the Demo FatAIMicrosoft’s 700B AI bet: Hype or a real retail crystal ball?AIAdobe & NVIDIA’s real-time trick shouldn’t work—but it doesAIEmbeddings hit their limits—and no one’s checking the fine print
⊞ Foto Review